Essays/Building

Speed is a structure

Speed in a business is not hurry. It comes from a structure built to run many feedback loops, and from judgment about when to push and when to slow down.

Speed in a business is a result. It comes from a structure built so that the whole company accelerates together when it is pushed, and from knowing when to push. At the marketing agency we chose speed because it was the only advantage open to us, and then we had to build what made it possible.

Why we chose speed

Before the agency had any success, we asked what could set us apart from the competition. We had no money, almost no resources, very little know-how and not much knowledge of the industry. Even calling ourselves a marketing agency was generous. Apart from me, most of the team had never worked in the field. It was a new team in a new domain.

We decided that the one way we could win was to move fast, faster than everyone else.

What is the difference between hurry and speed?

Deciding to be fast does nothing by itself. A person in a hurry rushes, makes mistakes, burns out and produces motion without progress. A fast machine has a structure underneath it, so that when it is pushed the whole thing accelerates together. We were trying to build the second.

What our machine was built around is in Invent your own game. We analyzed the logical structure of our competitors’ campaigns so that we could rebuild them technically. When a campaign did not work the first time, we iterated until it did.

Speed lives inside iteration. A loop is one pass from a guess to feedback from the market, and each pass turns an assumption into knowledge. The number of complete loops matters more than the speed of any single step. A team that runs one rough loop a week has twelve pieces of real feedback after three months. A team that runs one careful loop a month has three. Each loop also aims the next one better, so the lead grows. The checkpoints along the way are in Ship the prototype, then listen.

When we pressed the accelerator

At some point we saw a trend. The more campaigns we launched alongside our partners, the more stable and the higher our revenue became. No single campaign brought in a huge sum, only a decent one, and a few did well with almost no effort. Underneath it was a principle of probability. With enough attempts run through a structure that worked, some would hit, and we did not have to be brilliant every time.

When we understood that, we dropped everything else and concentrated on creating more campaigns. I took it further than I usually would. I upset some people in the company, especially on the operational and administrative side, because sometimes I did not speak to them for days and they had to make decisions on their own. They kept finding their own solutions without needing us, which turned out to be good for them. Finance and operations were out of balance for a while, because I did not have the time to grow those too. It was the right call.

A decision like that has to be made before being sure. Often there is only a hunch that pressing on one point will grow the whole business. By the time sales reach a level where it feels safe, the opportunity in the market may already have closed.

Once I knew the thing could work, I built a larger marketing team to take it over, and I moved on to other things. That handover is also how I tell a pattern from a fluke.

If the thing only works when you personally touch it, it is not yet a pattern, it is your talent, and talent does not scale.

When is it right to slow down?

Speed is not a constant. It matters most at the very start. With not enough money, no users and no feedback, the best move is to go as fast as possible toward a version that can get a first reaction, because early mistakes are almost always small and recoverable. After that, knowing when to push and when to ease off is learned over time. These are the signals I would name.

  • Slow down when the feedback stops making sense. Data that cannot be interpreted means the market has context the company is missing.
  • Slow down when a wrong move would be large and irreversible.
  • Speed up when the feedback is legible and the mistakes are small and recoverable.

I make one exception to the rule of going fast: an ordinary company should not race the technology itself. Enormous companies are building the AI models, and matching them would take a staggering amount of money. It is wiser to let the technology develop and use it. The same holds for the generic parts of a business. Tools and templates already exist for nearly all of them. Using them at the beginning saves money and leaves the time for the part that sets the company apart.

I call speed a structure because the other kind does not last. A founder who is fast by force of will gets a frantic, tired team that makes errors at speed. A structure built for speed gives a calm team that moves quickly because the machine underneath it was designed to.

Back to the essays

Field Notes

A letter from Andrei Hasna on building with AI agents

Subscribe by email

By clicking “Subscribe”, I confirm that I want to get email delivered to my address and agree to Andrei Hasna’s Privacy Policy and Terms.

In each letter

  • New essays on AI agents and the systems behind them
  • What Hasna is building, and why
  • Costs, routing and results from real work
  • What did not work
  • News of the book and the podcast